On Credibility of Medical Artificial Intelligence Technologies based on Physician-Patient Trust

Abstract

Since the 21st century, artificial intelligence (AI) technology has been widely used in the medical field, and has shown more accuracy and efficiency than human doctors for diagnosis. However, as the internal mechanisms are often unclear, an AI system is usually regarded as a "black box." This "black box" problem has raised concerns regarding the credibility of AI. However, one view claims that the concern regarding the "black box" is unnecessary, as human doctors do not seem not much better than AI in terms of providing diagnoses based on intuition and experience. In many cases, the human intuitions and/or experiences are also difficult to explain.Thus, if we can trust human doctors, it seems reasonable to trust medical AI. In this study, I attempt to respond this view, based on a theory of physician-patient trust.